A wide range of cellular proteins, both nuclear and cytoplasmic, are post-translationally modified by the addition of the monosaccharide 2-acetamido-2-deoxy-β-D-glucopyranoside (β-N-acetylglucosamine) which is attached via an O-glycosidic linkage.1 This modification is generally referred to as O-linked N-acetylglucosamine or O-GlcNAc. The enzyme responsible for post-translationally linking β-N-acetylglucosamine (GlcNAc) to specific serine and threonine residues of numerous nucleocytoplasmic proteins is O-GlcNAc transferase (OGT).2-5 A second enzyme, known as glycoprotein 2-acetamido-2-deoxy-β-D-glucopyranosidase (O-GlcNAcase)6,7 removes this post-translational modification to liberate proteins making the O-GlcNAc-modification a dynamic cycle occurring several times during the lifetime of a protein.8
O-GlcNAc-modified proteins regulate a wide range of vital cellular functions including, for example, transcription,9-12 proteasomal degradation,13 and cellular signaling.14 O-GlcNAc is also found on many structural proteins.15-17 For example, it has been found on a number of cytoskeletal proteins, including neurofilament proteins,18,19 synapsins,6,20 synapsin-specific clathrin assembly protein AP-3,7 and ankyrinG.14 O-GlcNAc modification has been found to be abundant in the braill.21,22 It has also been found on proteins clearly implicated in the etiology of several diseases including Alzheimer's disease (AD) and cancer.
For example, it is well established that AD and a number of related tauopathies including Downs' syndrome, Pick's disease, Niemann-Pick Type C disease, and amyotrophic lateral sclerosis (ALS) are characterized, in part, by the development of neurofibrillary tangles (NFTs). These NFTs are aggregates of paired helical filaments (PHFs) and are composed of an abnormal form of the cytoskeletal protein “tau”. Normally tau stabilizes a key cellular network of microtubules that is essential for distributing proteins and nutrients within neurons. In AD patients, however, tau becomes hyperphosphorylated, disrupting its normal functions, forming PHFs and ultimately aggregating to form NFTs. Six isoforms of tau are found in the human brain. In AD patients, all six isoforms of tau are found in NFTs, and all are markedly hyperphosphorylated.23,24 Tau in healthy brain tissue bears only 2 or 3 phosphate groups, whereas those found in the brains of AD patients bear, on average, 8 phosphate groups.25,26 A clear parallel between NFT levels in the brains of AD patients and the severity of dementia strongly supports a key role for tau dysfunction in AD.27,28 The precise causes of this hyperphosphorylation of tau remain elusive. Accordingly, considerable effort has been dedicated toward: a) elucidating the molecular physiological basis of tau hyperphosphorylation;29 and b) identifying strategies that could limit tau hyperphosphorylation in the hope that these might halt, or even reverse, the progression of Alzheimer's disease30-33 Thus far, several lines of evidence suggest that up-regulation of a number of kinases may be involved in hyperphosphorylation of tau,21,34,35 although very recently, an alternative basis for this hyperphosphorylation has been advanced.21 
In particular, it has emerged that phosphate levels of tau are regulated by the levels of O-GlcNAc on tau. The presence of O-GlcNAc on tau has stimulated studies that correlate O-GlcNAc levels with tau phosphorylation levels. The interest in this field stems from the observation that O-GlcNAc modification has been found to occur on many proteins at amino acid residues that are also known to be phosphorylated.36-38 Consistent with this observation, it has been found that increases in phosphorylation levels result in decreased O-GlcNAc levels and conversely, increased O-GlcNAc levels correlate with decreased phosphorylation levels.39 This reciprocal relationship between O-GlcNAc and phosphorylation has been termed the “Yin-Yang hypothesis”40 and has gained strong biochemical support by the discovery that the enzyme OGT4 forms a functional complex with phosphatases that act to remove phosphate groups from proteins.41 Like phosphorylation, O-GlcNAc is a dynamic modification that can be removed and reinstalled several times during the lifespan of a protein. Suggestively, the gene encoding O-GlcNAcase has been mapped to a chromosomal locus that is linked to AD.7,42 Hyperphosphorylated tau in human AD brains has markedly lower levels of O-GlcNAc than are found in healthy human brains.21 It has been shown that O-GlcNAc levels of soluble tau protein from human brains affected with AD are markedly lower than those from healthy brain.21 Furthermore, PHF from diseased brain was suggested to lack completely any O-GlcNAc modification whatsoever.21 The molecular basis of this hypoglycosylation of tau is not known, although it may stem from increased activity of kinases and/or dysfunction of one of the enzymes involved in processing O-GlcNAc. Supporting this latter view, in both PC-12 neuronal cells and in brain tissue sections from mice, a nonselective N-acetylglucosamindase inhibitor was used to increase tau O-GlcNAc levels, whereupon it was observed that phosphorylation levels decreased.21 The implication of these collective results is that by maintaining healthy O-GlcNAc levels in AD patients, such as by inhibiting the action of O-GlcNAcase, one should be able to block hyperphosphorylation of tau and all of the associated effects of tau hyperphosphorylation, including the formation of NFTs and downstream effects. However, because the proper functioning of the β-hexosaminidases is critical, any potential therapeutic intervention for the treatment of AD that blocks the action of O-GlcNAcase would have to avoid the concomitant inhibition of both hexosaminidases A and B.
Neurons do not store glucose and therefore the brain relies on glucose supplied by blood to maintain its essential metabolic functions. Notably, it has been shown that within brain, glucose uptake and metabolism decreases with aging.43 Within the brains of AD patients marked decreases in glucose utilization occur and are thought to be a potential cause of neurodegeneration.44 The basis for this decreased glucose supply in AD brain45-47 is thought to stem from any of decreased glucose transport,48,49 impaired insulin signaling,50,51 and decreased blood flow.52 
In light of this impaired glucose metabolism, it is worth noting that of all glucose entering into cells, 2-5% is shunted into the hexosamine biosynthetic pathway, thereby regulating cellular concentrations of the end product of this pathway, uridine diphosphate-N-acetylglucosamine (UDP-GleNAc).53 UDP-GlcNAc is a substrate of the nucleocytoplasmic enzyme O-GlcNAc transferase (OGT),2-5 which acts to post-translationally add GlcNAc to specific serine and threonine residues of numerous nucleocytoplasmic proteins. OGT recognizes many of its substrates54,55 and binding partners41,56 through its tetratricopeptide repeat (TPR) domains.57,58 As described above, O-GlcNAcase6,7 removes this post-translational modification to liberate proteins making the O-GlcNAc-modification a dynamic cycle occurring several times during the lifetime of a protein.8 O-GlcNAc has been found in several proteins on known phosphorylation sites,10,37,38,59 including tau and neurofilaments.60 Additionally, OGT shows unusual kinetic behaviour making it exquisitely sensitive to intracellular UDP-GlcNAc substrate concentrations and therefore glucose supply.41 
Consistent with the known properties of the hexosamine biosynthetic pathway, the enzymatic properties of OGT, and the reciprocal relationship between O-GlcNAc and phosphorylation, it has been shown that decreased glucose availability in brain leads to tau hyperphosphorylation.44 Therefore the gradual impairment of glucose transport and metabolism, whatever its causes, leads to decreased O-GlcNAc and hyperphosphorylation of tau (and other proteins). Accordingly, the inhibition of O-GlcNAcase should compensate for the age related impairment of glucose metabolism within the brains of health individuals as well as patients suffering from AD or related neurodegenerative diseases.
These results suggest that a malfunction in the mechanisms regulating tau O-GlcNAc levels may be vitally important in the formation of NFTs and associated neurodegeneration. Good support for blocking tau hyperphosphorylation as a therapeutically useful intervention61 comes from recent studies showing that when transgenic mice harbouring human tau are treated with kinase inhibitors, they do not develop typical motor defects33 and, in another case,32 show decreased levels of insoluble tau. These studies provide a clear link between lowering tau phosphorylation levels and alleviating AD-like behavioural symptoms in a murine model of this disease. Indeed, pharmacological modulation of tau hyperphosphorylation is widely recognized as a valid therapeutic strategy for treating AD and other neurodegenerative disorders.62 
Small-molecule O-GlcNAcase inhibitors, to limit tau hyperphosphorylation, have been considered for treatment of AD and related tauopathies.63 Specifically, the O-GlcNAcase inhibitor thiamet-G has been implicated in the reduction of tau phosphorylation in cultured PC-12 cells at pathologically relevant sites.63 Moreover, oral administration of thiamet-G to healthy Sprague-Dawley rats has been implicated in reduced phosphorylation of tau at Thr231, Ser396 and Ser422 in both rat cortex and hippocampus.63 
There is also a large body of evidence indicating that increased levels of O-GlcNAc protein modification provides protection against pathogenic effects of stress in cardiac tissue, including stress caused by ischemia, hemorrhage, hypervolemic shock, and calcium paradox. For example, activation of the hexosamine biosynthetic pathway (HBP) by administration of glucosamine has been demonstrated to exert a protective effect in animals models of ischemia/reperfiision,64-70 trauma hemorrhage,71-73 hypervolemic shock,74 and calcium paradox.64,75 Moreover, strong evidence indicates that these cardioprotective effects are mediated by elevated levels of protein O-GlcNAc modification.64,65,67,70,72,75-78 There is also evidence that the O-GlcNAc modification plays a role in a variety of neurodegenerative diseases, including Parkinson's disease and Huntington's disease.79 
Humans have three genes encoding enzymes that cleave terminal β-N-acetyl-glucosamine residues from glycoconjugates. The first of these encodes O-GlcNAcase. O-GlcNAcase is a member of family 84 of glycoside hydrolases that includes enzymes from organisms as diverse as prokaryotic pathogens to humans (for the family classification of glycoside hydrolases see Coutinho, P. M. & Henrissat, B. (1999) Carbohydrate-Active Enzymes server at URL: http://afmb.cnrs-mrs.fr/CAZY/.27,28 O-GlcNAcase acts to hydrolyse O-GlcNAc off of serine and threonine residues of post-translationally modified proteins.1,6,7,80,81 Consistent with the presence of O-GlcNAc on many intracellular proteins, the enzyme O-GlcNAcase appears to have a role in the etiology of several diseases including type II diabetes,14,82 AD,16,21,83 and cancer.22,84 Although O-GlcNAcase was likely isolated earlier on,18,19 about 20 years elapsed before its biochemical role in acting to cleave O-GlcNAc from serine and threonine residues of proteins was understood.6 More recently O-GlcNAcase has been cloned,7 partially characterized,20 and suggested to have additional activity as a histone acetyltransferase.20 However, little was known about the catalytic mechanism of this enzyme.
The other two genes, HEXA and HEXB, encode enzymes catalyzing the hydrolytic cleavage of terminal β-N-acetylglucosamine residues from glycoconjugates. The gene products of HEXA and HEXB predominantly yield two dimeric isozymes, hexosaminidase A and hexosaminidase B, respectively. Hexosaminidase A (αβ), a heterodimeric isozyme, is composed of an α- and a β-subunit. Hexosaminidase B (ββ), a homodimeric isozyme, is composed of two β-subunits. The two subunits, α- and β-, bear a high level of sequence identity. Both of these enzymes are classified as members of family 20 of glycoside hydrolases and are normally localized within lysosomes. The proper functioning of these lysosomal β-hexosaminidases is critical for human development, a fact that is underscored by the tragic genetic illnesses, Tay-Sach's and Sandhoff diseases which stem from a dysfunction in, respectively, hexosaminidase A and hexosaminidase B.85 These enzymatic deficiencies cause an accumulation of glycolipids and glycoconjugates in the lysosomes resulting in neurological impairment and deformation. The deleterious effects of accumulation of gangliosides at the organismal level are still being uncovered.86 
As a result of the biological importance of these β-N-acetyl-glucosaminidases, small molecule inhibitors of glycosidases87-90 have received a great deal of attention,91 both as tools for elucidating the role of these enzymes in biological processes and in developing potential therapeutic applications. The control of glycosidase function using small molecules offers several advantages over genetic knockout studies including the ability to rapidly vary doses or to entirely withdraw treatment.
However, a major challenge in developing inhibitors for blocking the function of mammalian glycosidases, including O-GlcNAcase, is the large number of functionally related enzymes present in tissues of higher eukaryotes. Accordingly, the use of non-selective inhibitors in studying the cellular and organismal physiological role of one particular enzyme is complicated because complex phenotypes arise from the concomitant inhibition of such functionally related enzymes. In the case of β-N-acetylglucosaminidases, many compounds that act to block O-GlcNAcase function are non-specific and act potently to inhibit the lysosomal β-hexosaminidases.
A few of the better characterized inhibitors of β-N-acetyl-glucosaminidases which have been used in studies of O-GlcNAc post-translational modification within both cells and tissues are streptozotocin (STZ), 2″-methyl-α-D-glucopyrano-[2,1-d]-Δ2′-thiazoline (NAG-thiazoline) and O-(2-acetamido-2-deoxy-D-glucopyranosylidene)amino N-phenylcarbamate (PUGNAc).14,92-95 
STZ has long been used as a diabetogenic compound because it has a particularly detrimental effect on β-islet cells.96 STZ exerts its cytotoxic effects through both the alkylation of cellular DNA96,97 as well as the generation of radical species including nitric oxide.98 The resulting DNA strand breakage promotes the activation of poly(ADP-ribose) polymerase (PARP)99 with the net effect of depleting cellular NAD+levels and, ultimately, leading to cell death.100,101 Other investigators have proposed instead that STZ toxicity is a consequence of the irreversible inhibition of O-GlcNAcase, which is highly expressed within β-islet cells.92,102 This hypothesis has, however, been brought into question by two independent research groups.103,104 Because cellular O-GlcNAc levels on proteins increase in response to many forms of cellular stress105 it seems possible that STZ results in increased O-GlcNAc-modification levels on proteins by inducing cellular stress rather than through any specific and direct action on O-GlcNAcase. Indeed, Hanover and coworkers have shown that STZ functions as a poor and somewhat selective inhibitor of O-GlcNAcase106 and although it has been proposed by others that STZ acts to irreversibly inhibit O-GlcNAcase,107 there has been no clear demonstration of this mode of action. More recently, it has been shown that STZ does not irreversibly inhibit O-GlcNAcase.108 
NAG-thiazoline has been found to be a potent inhibitor of family 20 hexosaminidases,90,109 and more recently, the family 84 O-GlcNAcases.108 Despite its potency, a downside to using NAG-thiazoline in a complex biological context is that it lacks selectivity and therefore perturbs multiple cellular processes.
PUGNAc is another compound that suffers from the same problem of lack of selectivity, yet has enjoyed use as an inhibitor of both human O-GlcNAcase6,110 and the family 20 human β-hexosaminidases.111 This molecule, developed by Vasella and coworkers, was found to be a potent competitive inhibitor of the β-N-acetyl-glucosaminidases from Canavalia ensiformis, Mucor rouxii, and the β-hexosaminidase from bovine kidney.88 It has been demonstrated that administration of PUGNAc in a rat model of trauma hemorrhage decreases circulating levels of the pro-inflammatory cytokines TNF-α and IL-6.112 It has also been shown that administration of PUGNAc in a cell-based model of lymphocyte activation decreases production of the cytokine IL-2.113 Subsequent studies have indicated that PUGNAc can be used in an animal model to reduce myocardial infarct size after left coronary artery occlusions.114 Of particular significance is the fact that elevation of O-GlcNAc levels by administration of PUGNAc, an inhibitor of O-GlcNAcase, in a rat model of trauma hemorrhage improves cardiac function.112,115 In addition, elevation of O-GlcNAc levels by treatment with PUGNAc in a cellular model of ischemia/reperfusion injury using neonatal rat ventricular myocytes improved cell viability and reduced necrosis and apoptosis compared to untreated cells.116 
More recently, it has been suggested that the selective O-GlcNAcase inhibitor NButGT exhibits protective activity in cell-based models of ischemia/reperfusion and cellular stresses, including oxidative stress.117 This study suggests the use of O-GlcNAcase inhibitors to elevate protein O-GlcNAc levels and thereby prevent the pathogenic effects of stress in cardiac tissue.
International patent applications PCT/CA2006/000300, filed 1 Mar. 2006, published under No. WO 2006/092049 on 8 Sep. 2006; PCT/CA2007/001554, filed 31 Aug. 2007, published under No. WO 2008/025170 on 6 Mar. 2008; PCT/CA2009/001087, filed 31 Jul. 2009, published under No. WO 2010/012106 on 4 Feb. 2010; PCT/CA2009/001088, filed 31 Jul. 2009, published under WO 2010/012107 on 4 Feb. 2010; and PCT/CA2009/001302, filed 16 Sep. 2009, published under WO 2010/037207 on 8 Apr. 2010, describe selective inhibitors of O-GlcNAcase.